Our results on axion searches using blazars signify an important step beyond the state of the art. On the one hand, the derived bounds are strongest to date derived from gamma-ray observations. On the other hand, our magnetic field model self-consistently describes the broad band emission of the blazar. For the first time, we also incorporated the uncertainty of the intervening magnetic fields in our statistical analysis, which renders our results more robust. Furthermore, our constraints on the intergalactic magnetic field improve previous bounds by a factor of two and combine data from different instruments self-consistently.
For the remaining time, my team and I will focus on the search for heavier axions that could decay into photons. Such a decay should be visible in background radiation fields that penetrate the universe. These background fields are not precisely known and we will model the astrophysical mechanisms in addition to the axion decay self-consistently and confront our predictions with data. Furthermore, we will conduct novel searches for axions produced in supernova explosions. In such explosions that occur at the end of the lifetime of massive stars, axions could be copiously produced and subsequently convert to gamma rays in the magnetic field of our Milky Way. As a result, we expect a short gamma-ray burst that we will look for in Fermi LAT data. We have already started to collect a sample of archival supernova explosions detected with optical telescopes that are well suited for such searches.
Regarding laboratory searches, my team will continue to characterize our single photon detector and improve the suppression of backgrounds. Our preliminary results on the efficiency together with the achieved background suppression mark an important step beyond the state of the art for such detectors. We will further investigate the response of the detector to photons at different wavelengths in order to calibrate its energy resolution. We will also work on the improvement of machine learning algorithms for signal and background discrimination. In particular, convolutional neural networks for time series classification offer an exciting opportunity in this regard. Lastly, we will develop an optical filter bench that can be operated within the cryostat that also houses the detector. Such a filter bench can in principle reject photons at the wrong wavelength but we will have to ensure a high transmission at the correct wavelength.
